1 / 21

Boosting

Explore common errors in midterm tests and grasp the essentials of boosting techniques, focusing on maintaining sanity checks and understanding key concepts like entropy, information gain, and decision tree principles. Learn about controlling overfitting and the nuances of Bayesian networks and graphical model inference in this comprehensive guide.

clydevictoria
Download Presentation

Boosting

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Boosting Recitation 10 OznurTastan

  2. Outline Overview of common mistakes in midterm Boosting

  3. Sanity checks Entropy for discrete variables is always non-negativeand equals zero only if the variable takes on a single value Information gain is always non-negative

  4. Sanity checks • In decision trees: You cannot obtain a leaf that has no training examples If a leaf contains examples from multiple classes, you predict the most common class. If there are multiple, you predict any of the most common classes.

  5. Common mistakes Many people only stated one of either of the problems.

  6. Common mistakes 6.3 Controlling overfitting Increase the number of training examples in logistic regression, the bias remains unchanged. MLE is an approximately unbiased estimator. 11 Bayesian networks’ 12 Graphical model inference Entries in potential tables aren't probabilities Many people forgot about the possibility of accidental independences.

  7. Boosting As opposed to bagging and random forest learn many big trees Learn many small trees (weak classifiers) Commonly used terms: Learner = Hypothesis = Classifier

  8. Boosting Given weak learner that can consistently classify the examples with error ≤1/2-γ A boosting algorithm can provably construct single classifier with error ≤ε where ε and γ are small.

  9. AdaBoost In the first round all examples are equally weighted D_t(i)=1/N At each run: Concentrate on the hardest ones: The examples that are misclassified in the previous run are weighted more so that the new learner focuses on them. At the end: Take a weighted majority vote.

  10. Formal description this is a distribution over examples this is the classifier or hypothesis weighted error mistake on an example with high weight costs much. weighted majority vote

  11. Updating the distribution Correctly predicted this example decrease the weight of the example Mistaken. Increase the weight of the example

  12. Updating Dt weighted error of the classifier

  13. 1/2 ln((1-0.3)/0.3)

  14. When final hypothesis is too complex

  15. Margin of the classifier

  16. Cumulative distribution of the margins Although the final classifier is getting larger, the margins are increasing.

  17. Advantages Fast Simple and easy to program No parameters to tune (except T) Provably effective • Performance depends on the data and the weak learner • Can fail if the weak learners are too complex (overfitting) • If the weak classifiers are too simple (underfitting)

  18. References Miroslav Dudik lecture slides.

More Related